/cell-fine-tune

cell fine tune

Primary LanguagePythonMIT LicenseMIT

cell-fine-tune

instruction

Installation

install the requirements:

pip install -r requirements.txt

Setup

Download all files (weights) from the folder final_weights from the google drive and put them in the folder (../)cell-fine-tune/weights
https://drive.google.com/drive/folders/1Ngp34zO8bx39z0ZSqwtCpZoCjRCEpc24?usp=sharing

Running

Run mask prediction:

python infer.py (--model <hybrid/pure>) (--image_folder <>) (--save_folder <>)

--model hybrid runs our Hybrid model

--model pure runs MaskRCNN

Evaluation

Basic statistics can be generated using the notebook: evaluation/evaluate_prediction.ipynb

Issues

In case of problems with libomp on Mac (pytorch/pytorch#20030), add:

brew install libomp

Based on

https://github.com/Confusezius/unet-lits-2d-pipeline
https://github.com/matterport/Mask_RCNN